Variability of angina symptoms over a 5-year period was examined in a prospective study, in which 7,109 British middle-aged men completed two chest pain questionnaires, Q1 (1978-1980) and Q5 (1983-1985), and were classified as having no chest pain, nonexertional chest pain, or angina (Q) (exertional chest pain) on each occasion. Within persons, there was considerable variability in response to the chest pain questions at Q1 and Q5. Angina (Q) persistence showed marked associations with previous myocardial infarction, diagnosed angina, electrocardiogram ischemia, and subsequent major ischemic heart disease events from Q5 onward. Compared with men without angina (Q), the age-adjusted hazard ratios were 1.5 (95% confidence interval (CI): 1.1, 2.2) for angina (Q) at Q1 only, 2.6 (95% CI: 2.1, 3.2) for angina (Q) at Q5 only, and 3.4 (95% CI: 2.8, 4.3) for angina (Q) on both occasions. For men without diagnosed ischemic heart disease, for whom apparent remission of angina (Q) was particularly frequent, a similar pattern of association was found between angina (Q) persistence and subsequent major events. In middle-aged men, exertional chest pain is a strong indicator of major coronary risk but frequently appears transient in the longer term. Persistently reported symptoms are associated with severe disease and a poor prognosis.